Dan Davies is a managing director at Frontline Analysts, a research firm. He is the author of Lying For Money and is also available on Substack.
Could it be . . . could it possibly be the case that a huge amount of economic punditry, comment and even policy was based on a simple data problem?
Well yes, obviously, that happens all the time. Now it appears that “expansionary bankruptcy” might be on a factual footing almost as weak as “expansionary austerity”: a new study by the European Systemic Risk Board casts significant doubt on whether so-called zombie companies are anything like as great a threat to the economy as was previously feared.
What is a zombie company? It’s annoyingly difficult to define in rigorous terms, so we’ll come back to that question. The underlying idea, though, is that it’s a lousy company that in some sense ought to have gone bankrupt, but which is continuing to pay wages and suppliers by taking out soft loans from an equally lousy bank that doesn’t want to admit that its client is bust.
It’s often assumed that there are lots of these dysfunctional bank/corporate relationships in Japan and Europe, and that as well as being an aesthetic offence against the values of creative destruction, they have a sclerotic effect on the economy as a whole. Various research papers over the years appear to have found that when you have lots of zombies hanging around, more vital and efficient companies are starved of funding.
This is, of course, a wholly different phenomenon from a lousy company which should never have existed in the first place but which is able to continue funding its losses by taking successive equity injections from venture capital firms who don’t want to admit that their investment is bust. That’s not a zombie, requires no investigation and certainly doesn’t have any crowding-out effects. Please don’t raise that question again.
All previous work, however, has been carried out using aggregate measures of one kind or another, correlating the prevalence of zombies in an industry with various measures of output and inferring causation. The ESRB team, however, have access to AnaCredit, the analytical credit data set of the ECB. This is a map equal in size to the territory — it includes literally every business loan in the euro area since 2018 that was over €25,000 in size. When combined with an equally huge database of company accounts, it’s possible to see whether zombies do actually get special treatment.
It turns out that they don’t. In general, zombie firms pay more for their credit than healthy companies, and are less likely to have new loans extended. Banks with a lot of zombie customers don’t seem to reduce their lending to healthy firms, and if anything they tend to reduce their pricing to their decent customers. If there is any negative effect of “zombification”, it has to work through much more circuitous channels; the zombies don’t starve healthy firms of finance.
Or alternatively, there might not be any effect to measure. In an Appendix, the authors find that in any given year, there’s only about a 30 per cent chance that a company identified as a zombie on their criteria will still be considered a zombie next year — more than two-thirds of them either recover or go bust in the next twelve months. This is because the ESRB team has used a quite restrictive definition based on negative return on assets (unprofitability), negative change in fixed assets (to screen out growth companies and startups) and ebitda/financial debt less than 5 per cent (to pick up financial distress). Other studies, using different measures have much higher persistency rates.
And this underlines the whole problem here — what you get out of the data is determined to a great extent by what you went looking for. Working with aggregate statistics is the professional deformation of empirical economics — like a geologist working with seismics, you have information that’s collected at a much lower resolution than the structure of the thing you’re hoping to understand, so you have to impose a theoretical framework on the data to draw a picture. That’s all fine, unless you forget you’re doing so, or get confused about whether the conclusions are coming from the data or from the framework you imposed.
The idea that it’s a good thing for the economy when a company goes bankrupt is attractive to some economists for theoretical reasons. It’s true in some circumstances and, because it’s got that attractive whiff of brimstone, policymakers can feel like they’re being tough and taking the decisions that softie politicians don’t have the guts for. But there’s an alternative theory — that insolvency is bad, and that productive enterprises ought to be given as much of a chance as possible before winding them up — which is also true in some circumstances. The American economy, with a very borrower-friendly bankruptcy code, does well out of a system which closes down companies quickly, but that doesn’t necessarily mean that copying US foreclosure practices will get you US outcomes.
And just as JK Galbraith defined the “acceptable rate of unemployment” as the rate that’s acceptable to those who have jobs, economists working for finance ministries and central banks don’t necessarily have as much skin in the game as you’d like when deciding what’s a temporarily troubled company and what’s a sclerotic basket case that needs to be taken to pieces for spare parts. If troubled European borrowers don’t affect credit availability for their healthy competitors, and have a better-than-even chance of recovering themselves, then there’s a disconcerting probability that what we’re looking at is not a zombie apocalypse, but rather a bunch of excitable policymakers who want to go round bashing their neighbours’ brains in with a shovel.
Source: Economy - ft.com
